. Statistical Literacy .



. Correlation (Association) & Causation .

One of the most commonly miss-interpreted ideas, especially in the media

Often a ‘correlation’ is given ‘causation’ link (or implied)

Want to live longer, walk faster
A firm handshake could be a sign of a longer life expectancy, according to British researchers.
Teenage sex 'leads to bad moods' in later life
Early Language Skills Reduce Preschool Tantrums, US news
Higher beer prices 'cut gonorrhoea rates' (link to BBC news)

‘Correlation’ words: associated, related, connected, linked, corresponding

Association keywords are taken to include associate, relate and predict:

- Tree-Lined Streets Associated With Lower Childhood Asthma Rates

- Kids’ problems related to drinking in pregnancy

- Smiles Predict Marriage Success; Finger Length Predicts Aggression in Men

‘Causation’ words: causes, blames, reason for, due to, because of, responsible for... Causation keywords are taken to include cause, effect and result.

-  Damp, Mouldy Homes May Cause Depression

-  Effects of child head injuries can last for years

. Language & Word choice .

Some words imply a negative outcome:

- Milk cancels health benefit of drinking tea: study

- Simple regimen eliminates chronic bad breath

- Cream with green tea extract hinders HIV transmission: study

- WHO: Sun Exposure Kills 60,000 Worldwide Each Year

Some have time-related keywords such as after, follow and leads:

- Fracture risk doubled after obesity surgery

- When toddlers point a lot, more words will follow

- Childhood Obesity Leads to Adolescent Obesity

Words which imply much, but assert little: such as more, less, many, lots, high, often, seldom

- British poll: Religion does more harm than good

- Dads More Likely Than Moms to Pass on MS

- Babies Born at Night More Likely to Die


Causal and Non-causal Language

Part 1: Listed below are sets of two "headlines" for the same study.

·  Sexual lyrics prompt teens to have sex

·  Listening to sexual lyrics associated with teen sex

·  Memory retention enhanced by sleep

·  People who sleep more, remember more!

·  Kids who take music lessons have bigger brains

·  Music lessons improve kids' brain development

Part 2: Determine whether each is causal or non-causal.

·  Daytime TV tied to poorer mental scores in elderly

·  Church attendance boosts immunity

·  Child anxiety linked to drug use

·  Political bias affects brain activity

·  Breakfast helps girls stay slim

·  Low self-esteem shrinks brain

·  Eat sweets, live longer

·  Disciplinarian parents have fat kids


. Support & Measurability .

What level of support does a claim have?

Is it repeatable? ‘Soft drinks could boost pancreatic cancer risk’

(no = weak support)

Is it measurable (before, during, after) or just observed?

Can a subject be assigned or not? ‘Weddings boost mood’

Repeatable & Assigning possible = strong support

‘flu vaccine reduces time off work’

. Think through the Context .

Look for any Obvious Errors

Look for ideas that are vague (especially words which imply much, but assert little)

Distinguish between Frequently & Likely

Look for the difference between Rates & Percentages

Look for Real Statistics and Speculative Statistics


. Practice .

Find statistical claim based articles in the media (newspaper, on-line)

1) What is the article about?

2) Is there any correlation or causation implied? What would you expect?

3) What words have been used relating to:

negative outcome

time-related keywords

Words which imply much, but assert little

What level of support does a claim have?

Is it repeatable, measurable & assigning possible

Look for any Obvious Errors

Look for ideas that are vague (especially words which imply much, but assert little)

Distinguish between Frequently & Likely

Look for the difference between Rates & Percentages

Look for Real Statistics and Speculative Statistics

Share, compare